AI-enabled innovation

2 min read

An academic debate about whether a machine can be named as an inventor must not discourage companies from investing in AI-enabled innovation and publishing the details of their inventions.

A Government-backed consultation, which closed at the end of November, has been seeking views from IP firms, innovators and other interested parties about whether a machine can be named as the inventor when seeking patent protection for an innovation.

The consultation was prompted by an earlier decision at the UK Intellectual Property Office in an important test case, known as DABUS. At the hearing, the Senior Hearing Officer commented that the ‘present patent system does not cater for inventions by AI’.

Patent attorneys are concerned that legislative changes in this area could hamper technological progress at a time of great promise for AI and machine learning.

Commenting Diego Black, partner and patent attorney at Withers & Rogers, said: “Existing UK and European legislation, backed by recent decisions, has made it clear to innovators that machines cannot be named as inventors in patent applications as the act requires the person to be limited to a natural person. Furthermore in AI and machine learning innovation there is always some element of human involvement or judgement in the process. To date, this clarity has served to promote investment in AI-enabled innovation in important areas, such as drug discovery and materials science.

“Any amendments would muddy the water and could force innovators to rethink the role of AI and machine learning as part of the innovation process. This would be a backward step and could discourage investment in AI-enabled R&D."

According to Black, “The problem hinges on IP rights and who owns them. Businesses pursue patent protection in order to share their ideas and benefit from a period of exclusivity, during which they are free to recoup some of their R&D investment. If an AI system or algorithm can be named as an inventor, which would be questionable on many levels, the use of an AI system could result in there being rights in the invention belonging to the AI. This could create uncertainty as to the ownership of the invention, especially in cases where the AI used in the invention process has been developed by a third party. This could undermine the company’s commercial rights and discourage investment.”

Having recently participated in the consultation, Withers & Rogers has confirmed its view that any invention will inevitably involve some element of human activity and that current UK and European patent law is fit for purpose. The firm has also emphasised that algorithms trained for use by AI-enabled systems, and the discoveries they generate, are patentable.

Black said, “This is an extremely exciting time for AI and machine learning technologies and we should not amend legislation that has already demonstrated its ability to facilitate investment.

“The role of AI in accelerating R&D processes by collecting and structuring large amounts of data is widely understood - not least for its role during the search for a vaccine for COVID-19 - but there is so much more to come. Already we are seeing AI-enabled systems playing a more integral role in research programmes by helping to define problems, optimise designs and come up with new possibilities. This important work must continue to attract investment, which means human beings must be recognised as the ‘source’ of any invention, not the technologies they develop to ‘aid’ or accelerate the R&D process.”

While asserting that the UK Patents Act 1977 is fit for purpose, patent attorneys at Withers & Rogers believe clarity and alignment with European legislation is important. The European Patent Office has recently introduced some helpful guidance by recommending that AI-enabled systems and algorithms are treated in the same way as software and consider AI to be a form of ‘super software’.

“The EPO’s decision to label AI-enabled innovations as ‘super software’ is extremely helpful as it underlines their patentability under existing legislation, and has helped provide certainty around the question of patentability of AI and machine learning based inventions. We would welcome a similar approach by the UKIPO,” Black concluded.